North Carolina puts out a net for tracking diseases
System relies on formats that were developed by CDC to transmit, track statewide data
- By Wilson P. Dizard III
- Mar 31, 2006
Just the fact that we can get data in a timely manner allows our epidemiologists to see things earlier and respond to them earlier.'
' Dr. Anna Waller, NC-DETECT
When Hurricane Katrina evacuees reached North Carolina last year, public health officials tracked the resulting surge in emergency room visits via a standards-driven statewide biosurveillance network.
'Katrina evacuees were using the emergency rooms for prescription refills and diabetes management,' said Dr. Anna Waller, principal investigator on the NC-DETECT project and a research associate professor at the University of North Carolina's emergency medicine department.
The project, also known as the North Carolina Disease Event Tracking and Epidemiological Collection Tool, is an example of the kind of biosurveillance system that the federal Health and Human Services Department is looking for as part of its early use of health IT.
NC-DETECT is the syndromic surveillance arm of the state's Public Health Information Network. It's designed to be the canary in the coal mine, providing early alerts of public health problems to providers and health care planners.
Its development underscores some of the challenges health officials face, such as inconsistent data formats in emergency rooms and disparate systems within hospitals. But it also is a model of progress.
'Utilizing NC-DETECT following hurricanes hitting North Carolina is timely and efficient compared to the emergency department paper log review and abstraction used in the past,' Waller said at a Health Information Management Systems Society meeting.
North Carolina public health professionals have used NC-DETECT to track the state's flu season. It can provide information on occupational health risks. And it is designed to detect bioterrorism and events such as a potential avian flu outbreak.
NC-DETECT includes data collected by the North Carolina Hospital Association in accord with a state mandate requiring hospitals to transmit selected emergency department data elements to the state at least daily.
The system is the primary statewide repository for emergency department data, funneling data into the statewide Public Health Information Network.
NC-DETECT relies on several standards incorporated in the Centers for Disease Control and Prevention's network architecture for health IT systems, known as the Public Health Information Network.Messaging system
In many cases, the system uses hospital data formatted into the Health Level 7 messaging system standard. However, as a messaging system, HL7 was not designed to transmit comprehensive medical data, according to Waller.
For example, not every data element captured in the hospital systems is available in the HL7 standard.
As a result, she said, many hospitals have chosen to submit their data to NC-DETECT as flat files.
The system uses secure File Transfer Protocol to receive the hospital data and downloads an aggregated file from submitting hospitals every 12 hours.
NC-DETECT then standardizes data according to the Data Elements for Emergency Department Systems 1.0 standard developed by CDC.
When patients enter emergency rooms, the hospital staff gathers so-called 'chief complaint' information on the spot. The chief complaint information can reflect local slang for health problems that varies across the state, Waller said.
NC-DETECT uses a tool called Emergency Medicine Text Pro- cessor to map free-text chief complaint data, such as transcribed patient statements, to standardized medical terms.
To help detect potential disease outbreaks, NC-DETECT uses a version of the Early Abberation Reporting System that CDC developed based on software from SAS Institute Inc. of Cary, N.C. One of the central functions of the EARS software, which the federal center provided for free, is to detect statistically significant increases in reported health syndromes.
'One of the major benefits is that we are collecting the data electronically in a timely manner, getting the data twice a day rather than four times a year,' Waller said. 'Just the fact that we can get data in a timely manner allows our epidemiologists to see things earlier and respond to them earlier.'
In addition to receiving emergency department data, NC-DETECT gathers data from the Carolinas Poison Center, the Prehospital Medical Information System, the Piedmont Wildlife Center and the North Carolina State College of Veterinary Medicine Laboratories.
One of the key challenges Waller and her team of systems and public health experts face has been monitoring the quality of the data NC-DETECT receives and how changes in hospital IT systems reflect the collected data elements.
'Emergency room data is notoriously dirty,' Waller said. The NC-DETECT staff constantly monitors the quality of the data and has at times removed hospitals from the data stream when their data became particularly unreliable.
Amy Ising, technical team director for NC-DETECT, emphasized the importance of metadata in pinpointing data quality problems.
'If we didn't have the basic metadata, we wouldn't know what to do [with the reports from the hospitals],' Ising said.
'For example, it now shows the average number of emergency department visits,' Ising said. 'If they usually say they get 50 a day but now are only getting 30, we can tell them some data is being dropped. The metadata can allow us to flag that something is not right, like a hospital sending a lot of duplicate visits.'
NC-DETECT's system for gathering public health data is part of a shift away from legacy public health data collection methods, in which a caregiver who encountered a patient with a reportable illness filled out a postcard and mailed it to the state capital.
Rather than focusing on specific reportable diseases, NC-DETECT groups infection-related emergency department visits into six syndromes referred to by terms such as influenza-like illness, fever/rash illness, neurological illness and respiratory illness.
NC-DETECT managers expect to expand the system to all 113 major hospitals in the state and are gearing up to process 4 million emergency department visits annually. The system already has processed more than 2.4 million visits by more than 1.2 million patients.
Project leaders expect to refine NC-DETECT by migrating it to a Java platform by the end of the year. The change will let system managers improve the application's online reporting function.
The system's main users are public health professionals. Because they have not used emergency department data until recently, they in many cases have difficulty understanding it or explaining the type of data they need, Waller said. Project leaders have made a point of involving stakeholders in system design and changes.
At the other end of the data pipeline, hospital IT professionals often are not familiar with emergency department procedures or the types of data that emergency departments generate. Moreover, hospitals themselves have not yet adopted interoperable IT systems.
Hospitals have focused chiefly on integrating their own IT systems rather than making them interoperable, according to public health IT specialists.
JoLynn Montgomery, research investigator at the University of Michigan's Epidemiology Department, echoed the NC-DETECT concerns.
'Hospital data systems are uniformly not uniform,' Montgomery said. 'Changing them over to a uniform system is not going to happen soon.'
Montgomery added, 'In terms of data standards, there is an effort [to adopt them] but it is not going to happen soon. We have come a long way, but we have a long way to go.'